CN114926611A - Holographic navigation scene graph knowledge inference method and device based on ontology - Google Patents
Holographic navigation scene graph knowledge inference method and device based on ontology Download PDFInfo
- Publication number
- CN114926611A CN114926611A CN202210597124.7A CN202210597124A CN114926611A CN 114926611 A CN114926611 A CN 114926611A CN 202210597124 A CN202210597124 A CN 202210597124A CN 114926611 A CN114926611 A CN 114926611A
- Authority
- CN
- China
- Prior art keywords
- behavior
- relation
- ship
- scene
- time
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T19/00—Manipulating 3D models or images for computer graphics
- G06T19/003—Navigation within 3D models or images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/22—Indexing; Data structures therefor; Storage structures
- G06F16/2291—User-Defined Types; Storage management thereof
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/26—Visual data mining; Browsing structured data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Computer Graphics (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a holographic navigation scene graph knowledge inference method and a device based on a body, which divide the behaviors of traffic objects into 3 levels of behaviors including microcosmic, mesoscopic and macroscopic; carrying out structured modeling on the behaviors of the traffic objects, defining and defining the concepts of the behaviors, the processes, the events and the like of the traffic objects in the ontology; using a navigation scene ontology knowledge base and adopting a relation calculation module to represent the interaction between the object and the behavior in the sub-scene; representing the change of the whole macroscopic scene through an event trigger mechanism, thereby reasoning the knowledge of the relationship between objects in the scene, behavior recognition, scene evolution and the like; the navigation scene graph is automatically subjected to knowledge reasoning. According to the invention, through systematic and definite semantic definition and display expression of ship behaviors, a sailor, a pilot and a VTS operator can accurately understand the faced traffic scene, and the problem of poor logical reasoning capability of the water traffic scene modeling method is solved.
Description
Technical Field
The invention belongs to the technical field of scene knowledge inference, and particularly relates to a holographic navigation scene graph knowledge inference method and device based on an ontology.
Background
The current inference method of traffic scenes is mainly based on probabilistic inference and is common in the field of road traffic. Plath et al propose a probabilistic inference model based on scene decomposition, which divides a driving scene into a plurality of sub-scenes according to a motion constraint relationship between entities, and then performs quantitative inference on each sub-scene using a bayesian network model (Platho M, et al, 2012, 2013).
In the field of water traffic, scene knowledge reasoning research mainly focuses on the estimation and prediction of ship behaviors. In 2017 to 2018, Vouros G A and the like propose a datAcron ontology model by relying on the European datAcron project. The datAcron model is centered on track fragment semantic class (track Part), links events (Event), Time (Time), geographic information (geotry), Moving objects (Moving Object), etc., and divides events into low-level events (including move, stop, turn, etc.) and high-level events (fishing, drifting, etc. in maritime applications). The model compresses the track by adopting a data compression technology, reserves key points of the track, and then semantically labels the track by using a datAcron body. Although scholars research identification methods of ship behaviors, the methods lack clear semantic definition and display expression for a ship behavior system and cannot enable sailors, pilots and VTS operators to accurately understand traffic scenes faced by the sailors, pilots and VTS operators.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: the holographic navigation scene graph knowledge reasoning method and device based on the ontology are used for achieving the function of automatic knowledge reasoning on the navigation scene graph.
The technical scheme adopted by the invention for solving the technical problems is as follows: a holographic navigation scene graph knowledge inference method based on an ontology comprises the following steps:
s0: the method comprises the steps of setting up a holographic navigation scene graph knowledge inference device based on a body, wherein the holographic navigation scene graph knowledge inference device comprises a data preprocessing module, a scene knowledge base storage module, an object behavior information extraction module, a relation calculation module, an event recognition module and an inference knowledge output module;
s1: according to the understanding of a driver on objects and behaviors in a navigation scene, the space-time behaviors of the traffic objects are divided into behaviors, processes and events including the traffic objects by combining the space-time scales of ship behaviors, semantic multi-scale modeling of the space-time behaviors of the traffic objects is carried out, and semantic features of the space-time behaviors of the traffic objects under different space-time scales are respectively represented; according to the motion characteristics, the self attribute characteristics and the topological characteristics of the environment of the traffic object, respectively defining the hierarchical relationship of the time-space behaviors of the traffic object as a microscopic behavior, a mesoscopic behavior and a macroscopic behavior; in semantic cognition, the behavior of a traffic object corresponds to a microscopic behavior, the process corresponds to a mesoscopic behavior, and the event corresponds to a macroscopic behavior;
s2: carrying out structured ontology modeling on the space-time behaviors of the traffic objects, and defining concepts and attributes of the traffic objects, including behaviors, processes and events, in the ontology model;
s3: forming a scene ontology knowledge base according to the ontology model, and representing the interaction of the space-time behaviors of the objects in the sub-scene by adopting a relation calculation module;
s4: acquiring static data, dynamic data and environmental data of an object related to a scene and inputting the static data, the dynamic data and the environmental data into a scene ontology knowledge base; extracting behavior information and process information of the traffic object according to whether the behavior of the object in the two time slices changes, wherein the behavior in the behavior information does not change and the behavior in the process information changes; performing semantic aggregation on a plurality of processes to obtain an event;
s5: and (3) representing the change of the whole macroscopic scene through an event trigger mechanism, and reasoning and outputting the evolved scene knowledge.
According to the scheme, in the step S1, the microscopic behaviors represent the change of the attributes and the relations of the traffic objects in the space-time dimension, and are basic behavior units for describing the traffic objects; the mesoscopic behaviors represent the behaviors of the traffic object in a larger time range and space range, and are the aggregation of microscopic behaviors; macroscopic behavior represents the behavior of traffic objects over a large temporal and spatial extent; the behaviors of the traffic object are unchanged from a certain scale, and all the behaviors represent macroscopic behaviors; the scale is the critical scale for macroscopic behavior.
According to the scheme, in the step S1, the semantic multi-scale modeling of the spatiotemporal behavior of the traffic object includes the following specific steps: modeling behaviors as changes of attribute elements and relation elements of the traffic object in time and space dimensions, including motion characteristic attribute changes, self characteristic attribute changes and topological relation changes; modeling the process as a persistence of behavior over a period of time with the object attribute behavior and the relationship behavior unchanged; an event is modeled as the evolution of the behavior and processes of a traffic object and its logical and temporal relationships over space-time, the event comprising at least two processes.
According to the scheme, in the step S2, the specific steps are as follows:
s21: let Object be the subject Object of space-time behavior occurrence, Object Attribution Object being a property element of a subject Object Relation Is a relationship element of the subject Object, f (Object) Attribution ,Object Relation ) Is a comprehensive expression of the attribute and the relationship of the subject object, t is the time when the space-time Behavior occurs, Behavior is the ith Behavior in the space-time object Behavior, Process is the jth Process in the space-time object Behavior, Event k Is k processes in the spatio-temporal object behavior; the ontological concepts of the behavior, processes and events of the traffic object are modeled as follows:
the behavior is
The process is that
The event is
Behavioral cognitive expression is:
Cog＝{o,b,t,p}；
s22: defining an ontology attribute element of the traffic object, wherein the ontology attribute element comprises an object attribute and a data attribute;
the object attribute is used for representing the relation between the classes in the ontology and is the key for recognizing the ship behavior; the object properties include:
HasTraj represents the attribution relationship between the ship and the track, Domain is ship, and Ranges is track;
the compises represents the inclusion relationship between track segments and sub-tracks, Domain is track, and Ranges is track;
occurs represents the occurrence time of a ship track segment, and comprises a sub-attribute occusBegin representing the start time of the track segment, and a sub-attribute occusEnd representing the end time of the track segment;
the Reflect time indicates the time of the ship behavior, and comprises a child attribute reflectBegin which represents the starting time of the behavior, and a child attribute reflectEnd which represents the ending time of the behavior;
because the behavior and the track have a one-to-one corresponding relation, the reflectBegin is consistent with the start time occusBegin of the track section corresponding to the behavior, the reflectEnd is consistent with the end time occusEnd of the track section corresponding to the behavior, and the characteristic is subjected to knowledge expansion in SWRL;
before and after represent the precedence relationship of instant, and the two are in reverse of relationship; the two attributes form the basis of a time logic expression mechanism, and the complex time relation in the interval is evolved based on the simple logic relation between moments;
hasBehavior represents the relationship between the ship and the behavior, Domain is ship, and Ranges is behavior;
hasTopo represents the spatial topological relation between the ship track and the navigation environment, and comprises 4 relations of point-line, point-plane, line-line and line-plane; domain is trajectory, Range is trafficRule;
follow represents the continuous relationship between ship tracks and is used for constructing the logical relationship between track segments;
reflexes represent the mapping relation between track segments and behaviors, Domain is track, Ranges is behavior, and is in inverse relation with isrefleclectedBy;
HasPoint represents the dependency relationship between a track segment and a track Point, Domain is track, range is Point, and the attributes include sub-attributes hasBeginPoint and hasEndPoint;
the data attribute represents the attribute of the ontology concept and is used for describing the state of the class in the ontology; the data attributes include:
HasSpeed represents the speed attribute of the ship;
InXSDDateTimeStamp represents UTC time of instant under time class;
the MMSI is an identification code of the ship and is used for uniquely determining the identity of the ship;
ShipType indicates the ship's category attribute.
According to the above scheme, in step S3, the topological relation, the orientation relation, the distance relation, the time relation, and the semantic relation between the object and the environment, and between the object and the object are calculated according to the requirements of the scene application, so as to obtain the knowledge of the process including the object, and the specific steps are as follows:
s31: the method comprises the following steps of calculating a topological relation for acquiring topological behavior knowledge between a traffic object and an environment, and specifically comprises the following steps:
s311: the ship is regarded as a mass point, and the ship motion track comprises ship motion track points;
s312: regarding traffic rules or infrastructure in a navigation environment of a ship as an entity in a two-dimensional space, wherein the entity comprises a point P, a line L and a plane A; performing mathematical modeling on the entity according to the spatial characteristics of the entity to form mathematical expressions comprising point positions, line segments, broken lines, circular areas, elliptical areas and polygonal areas;
s313: constructing a common ship topological behavior set comprising point-line, point-plane, line-line and line-plane according to a topological relation formed between a ship track and a navigation environment by combining a 9-intersection model DE-9IM based on dimension expansion; representing a topological behavior of the ship in the sailing environment as one of a set of topologies;
s32: the orientation relation is calculated for acquiring the orientation relation knowledge between the objects and the environment, and the method specifically comprises the following steps:
s321: constructing a ship-following motion coordinate system, taking a ship head line as the positive direction of an x axis, and taking a ship positive transverse line as the positive direction of a y axis;
s322: the x axis and the y axis are intersected to form four directional areas, and the relative azimuth angle is used for quantization from the positive transverse direction to the head-tail direction;
s33: calculating a distance relation for describing the distance between two ship objects, wherein the distance relation comprises quantitative description and qualitative description;
let D denote the distance between vessel A and vessel B, (x) A ，y A )，(x B ，y B ) Respectively representing the position coordinates of the ship A and the ship B, the quantitative description refers to the Euclidean distance between two ship objects:
the qualitative description is the threshold judgment of Euclidean distance through experience and cognitive level of a crew; let D t Is a qualitative description of D, D n 、D m 、D l Is a threshold defined for the distance between ship objects; when the azimuth and velocity of the two vessels are satisfied to constitute a collision risk, the orientation of the distance relationship between the two vessels is described as:
s34: calculating a time relationship for obtaining knowledge of the time relationship between the objects; the time relation is the description of the behavior and the event of the ship on the time dimension, and the time relation is divided into a time point and a time period according to the difference of the time dimension; the time point describes a certain moment or moment; the time period describes a time range;
in COLREGs, the time point relation and the time period relation include earlier than before t 1 、before(t 1 ,t 2 ) Later than after ter t 2 、after(t 1 ,t 2 ) Between Between (t) 1 ,t 2 ) Beginning with beginnWitht 1 Ending with endWitht 2 ；
S35: calculating semantic relations for obtaining semantic relation knowledge between objects, including describing relations between concepts, concepts and instances, the specific steps are:
s351: defining element concepts including ship, people and environment elements and action event element concepts in a water traffic scene;
s352: according to expert knowledge, a resource description framework is utilized to describe the semantic relation into a triple structure < main, predicate and object >, a definition domain, a value domain and a relation of the semantic relation and corresponding description are defined, and an entity element semantic relation knowledge base is formed; the method comprises the following steps:
the definition domain is a ship, the relation is happen, the value domain is an event element, and the event element is described as an event occurring on a certain ship;
the definition domain is a ship, the relation is execute, the value domain is a behavior element, and the behavior element is described as a certain ship executing a certain behavior;
the definition domain is a ship, the relation is hasType, the value domain is a parameter attribute, and the value domain is described as the type of a certain ship;
the definition domain is a ship, the relation is a hash, the value domain is a parameter attribute and is described as the name of a certain ship;
the definition domain is a ship, the relation is hassped, the value domain is a parameter attribute, and the value domain is described as the speed of a certain ship;
the defined domain is a ship, the relation is hascourse, the value domain is a parameter attribute and is described as the course of a certain ship;
the definition domain is a ship, the relation is hazize, the value domain is a parameter attribute, and the value domain is described as the size of a certain ship;
defining a domain as a behavior element, a relation as trigger, and a value domain as a behavior element, wherein the value domain is described as a behavior element triggering a certain behavior;
the definition domain is a behavior element, the relation is cause, the value domain is an event element, and the definition domain is described as an event caused by a certain behavior;
the definition domain is a behavior element, the relation is operatedBy, the value domain is an operator, and the definition domain is described as that a certain behavior is operated by a certain person;
defining a domain as an event element, a relationship as hasType, a value domain as a parameter attribute and describing the type of a certain event;
defining a domain as an event element, a relationship as trigger, and a value domain as a behavior element, wherein the value domain is described as a behavior element for triggering a certain behavior for a certain event;
the definition domain is an event element, the relation is cause, the value domain is an event element and is described as an event causing the event;
s353: and acquiring attributes corresponding to the entity concepts, and inputting the attributes into the element semantic relation knowledge base for relation matching to acquire corresponding semantic relation knowledge.
According to the above solution, in step S5, the scene knowledge includes the element type, attribute, and composition of the scene, the relationships between objects in the scene and between objects, and between objects and the scene, and the behavior, event, and scene evolution process of the objects in the scene; in the scene evolution process type, the included elements include a process type and a name, a process semantic description, a time period, an included object, an included behavior, a behavior relation set, a process-process relation, a process-scene relation and a process-element relation; in the scene element type, the included elements include element names, time, spatial features, attribute features, semantic descriptions, element relationships, element participation processes, element participation behaviors, element affiliated scenes, element-process relationships and element-scene relationships; in the behavior type, the included elements comprise the behavior type, the name, the behavior participant, the time, the place, the behavior precursor, the behavior successor and the participant relationship; in the state type, the included elements comprise the state type, the name, the time, the place, the state characteristics of the elements and the state of the scene; the state characteristics of the elements comprise positions, attributes and semantics; the state of the scene includes the number of elements, spatial distribution, and relationships between elements.
A holographic navigation scene graph knowledge inference device based on an ontology comprises a data preprocessing module, a scene knowledge base storage module, an object behavior information extraction module, a relation calculation module, an event recognition module and an inference knowledge output module; the data preprocessing module is used for preprocessing the data of the sensed full-factor information and providing instantiated available information for the ontology model; the scene knowledge base storage module is used for storing a holographic navigation scene graph knowledge base constructed based on the ontology; the object behavior information extraction module is used for extracting behavior information knowledge related to the motion characteristics, the attribute characteristics and the relation change among the objects from the acquired full element information by combining a scene knowledge base; the relation calculation module is used for calculating the related relations such as the topological relation, the orientation relation, the distance relation, the time relation, the semantic relation and the like between the objects and the environment and between the objects, so as to obtain the knowledge including the activities of the objects; the event identification module is used for acquiring semantic aggregation of activity knowledge and identifying sub-events occurring in a scene; and the inference knowledge output module is used for outputting the scene evolution knowledge after the event is triggered.
Furthermore, prot g is used as a rule ontology modeling tool, and a pellet inference machine is used as an inference knowledge output module; the ontology modeling tool prot g and the pellet inference machine both support OWL language and SWRL language; the OWL language is used for describing rule knowledge; the SWRL language is a regular knowledge inference language.
A computer storage medium having stored therein a computer program executable by a computer processor, the computer program executing a method of ontology-based holographic navigation scene map knowledge inference.
The beneficial effects of the invention are as follows:
1. the invention relates to a holographic navigation scene graph knowledge inference method and a device based on a body, which divide the behaviors of a traffic object into 3 levels of behaviors including microcosmic, mesoscopic and macroscopic; carrying out structural modeling on the space-time behaviors of the traffic objects, defining and defining the concepts of the behaviors, the processes, the events and the like of the traffic objects in an ontology; using a navigation scene ontology knowledge base and adopting a relation calculation module to represent the interaction between the object and the behavior in the sub-scene; representing the change of the whole macroscopic scene through an event trigger mechanism so as to deduce the knowledge of scene evolution; the navigation scene graph is automatically subjected to knowledge reasoning.
2. According to the invention, through systematic and definite semantic definition and display expression of ship behaviors, a sailor, a pilot and a VTS operator accurately understand the faced traffic scene, and the problem of poor logical reasoning capability of the water traffic scene modeling method is solved.
Drawings
FIG. 1 is a flow chart of an embodiment of the present invention.
FIG. 2 is an example diagram of an evolution of a navigation scene based on an event trigger mechanism according to an embodiment of the present invention.
Fig. 3 is a functional block diagram of an embodiment of the present invention.
FIG. 4 is a spatial topological relationship diagram of a ship and a navigation environment according to an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and specific embodiments.
Referring to fig. 1, the holographic navigation scene map knowledge inference method based on ontology includes the following steps:
step 1: according to the understanding of a driver on the objects and behaviors in the navigation scene, the evolution of the holographic navigation scene is divided into the behaviors, the processes and the changes of events of the water traffic objects by combining the space-time scale of the ship behaviors;
step 2: according to the behaviors, processes and events in the step 1, the concepts and the attributes of the behaviors, the processes and the events are determined, the hierarchical relations of the behaviors, the processes and the events of the water traffic object are respectively defined as the behaviors of a microscopic level, a mesoscopic level and a macroscopic level, and the definitions are given in the body modeling;
and 3, step 3: according to the ontology modeling method in the step 2, on the basis of forming an ontology knowledge base, a relational computation module is adopted to represent the interaction between the object and the behavior in the sub-scene; and characterizing the change of the whole macroscopic scene through an event trigger mechanism.
The technical scheme of the embodiment of the invention is as follows:
first, construction method
And defining the behavior of the traffic object. The change of the object behaviors is the core of the evolution of a navigation scene, and the time-space behaviors of the traffic object are divided into 3 behaviors of micro behaviors, mesoscopic behaviors and macroscopic behaviors based on the motion characteristics, the self attribute characteristics and the topological characteristics of the environment of the traffic object by combining the cognitive habits of human beings. Microscopic behavior: the behavior unit represents the behavior of the traffic object in a short-time local area and describes the basis of the traffic object. Mesoscopic behaviour: representing the behavior of a traffic object over a large temporal and spatial extent, is an aggregation of microscopic behaviors. In the mesoscopic behavior, the traffic object behavior indicates that the traffic object trajectory and the topological state of the navigation environment remain unchanged in the current spatiotemporal scale, for example, in a deceleration stage before the traffic object is ready to be anchored, the traffic object performs a plurality of continuous operations which may include acceleration, deceleration, steering and the like. Macroscopic behavior: the method represents the behaviors of the traffic object in a large time and space range, the behaviors of the traffic object are unchanged from a certain scale, and all the behaviors represent macroscopic behaviors, and the scale is the critical scale of the macroscopic behaviors. For example, for the whole trajectory of "berthing-debarking" of the traffic object near the berth, since the motion state and the topology state of the traffic object are changed in the process, the behavior can be regarded as a macroscopic behavior, and the macroscopic behavior at least comprises two mesoscopic behaviors or a combination of one microscopic behavior and the mesoscopic behavior.
Semantic multi-scale modeling of traffic object behavior. Based on the multi-scale characteristics of the space-time behaviors of the traffic objects, the traffic object behaviors are divided into 3 layers of behaviors, processes and events from the aspect of semantic modeling, and the semantic characteristics of the traffic object trajectories under different space-time scales are represented respectively. The behavior is modeled into a physical quantity for maintaining the motion characteristic state of the traffic object in the process of sailing, and the physical quantity is a semantic behavior reflected by the motion characteristic of the track of the traffic object. The process is modeled as the basic behavior which lasts for a period of time, namely the behavior which is generated between the same motion characteristic track points recorded by the track (recorded by AIS, GPS, radar and other means) of the traffic object is the behavior which maintains the second-order motion characteristic (the combined characteristic of the speed and the course) of the traffic object unchanged. Events are modeled as the evolution of the behavior and processes of traffic objects and their logical and temporal relationships over space-time. In semantic cognition, the behaviors and activities of the traffic objects correspond to microscopic behaviors; the process corresponds to mesoscopic behavior; events correspond to macroscopic behaviors.
Ontology modeling of behaviors, processes, and events. The ontology modeling of the object is the same as the method for constructing the holographic navigation scene graph based on the ontology. The ontological concept of behaviors, processes and events is modeled as follows:
behavior:
the process is as follows:
event:
object represents a subject Object in which spatiotemporal behavior occurs, Object Attribution Attribute element, Object, representing subject Object Relation Is a relationship element of the subject Object, f (Object) Attribution ,Object Relation ) Is the comprehensive expression of the attribute and the relation of the main object, t represents the time of the occurrence of the space-time Behavior, Behavior is the ith Behavior in the space-time object Behavior, Process is the jth Process in the space-time object Behavior, Event k Are k processes in spatiotemporal object behavior.
And defining the attribute of the ontology. The object attribute is used for representing the relation between classes in the ontology and is the key for recognizing the ship behavior. Examples of object properties in an ontology are as follows:
HasTraj represents the affiliation between the ship and the track, Domain is ship and Ranges is track.
The relations between track segments and sub-tracks are shown, Domain is track, and Ranges is track.
Occurs represents the occurrence time of a ship track segment, and comprises sub-attributes occusBegin and occusEnd which respectively represent the starting time and the ending time of the track segment.
Reflect time indicates the time when the ship behavior occurs, the child attribute refleBegin represents the starting time when the behavior occurs, and reflectEnd represents the ending time of the behavior. Because the behavior and the track have a one-to-one correspondence relationship, the reflexedbegin is consistent with the start time occusbegin of the track segment corresponding to the behavior, and the reflexend is consistent with the end time occusend of the track segment corresponding to the behavior, which is characterized by performing knowledge expansion in the SWRL.
before/after represents the precedence relationship of time instant and the two are in inverse relationship. The two attributes form the basis of a time logic expression mechanism, and the complex time relation in the interval is evolved based on the simple logic relation between moments.
hasBehavior, representing the relationship between ships and behaviors, Domain is ship and Ranges is behavior.
hasTopo represents the spatial topological relation between the ship track and the navigation environment, and comprises 4 relations: point-line, point-plane, line-line and line-plane. Domain is track and Range is track and RuffecRule.
Follow represents the continuous relationship between the ship tracks to construct the logical relationship between the track segments.
Reflets: the mapping relationship between the track segments and the behaviors is shown, Domain is track, Ranges is behavior, and isrefleclecytby is inverse (inverse of) relationship.
HasPoint: and showing the subordination relation between the track segment and the track Point, wherein Domain is track, and Range is Point, and the subordination relation comprises two sub-attributes, namely hasBeginPoint and hasEndPoint.
Data attributes represent attributes of an ontology concept that can be used to describe the state of classes in an ontology.
HasSpeed representing the speed attribute of a ship
InXSDDateTimeStamp representing UTC time of instant under time class
The MMSI is the identification code of the ship and can uniquely determine the identity of the ship.
ShipType indicates the class attribute of the ship.
And (5) performing relational operation. According to the requirements of scene application, the related relations such as topological relation, orientation relation, distance relation, time relation, semantic relation and the like between the objects and the environment and between the objects are calculated, so that the knowledge of the activities and the like of the objects is obtained.
(1) And (3) operation of topological relation: the method is used for acquiring the knowledge of the topological behavior between the traffic object and the environment. The method comprises the following steps: 1) the ship is regarded as a mass point, and the ship motion track is composed of ship motion track points. 2) The traffic rules or infrastructure in the navigation environment of the ship are regarded as entities in a two-dimensional space, and the entities comprise a point P, a line L and a plane A; mathematical modeling of the entity is performed based on the spatial characteristics of the entity, forming mathematical representations of point locations, line segments, polylines, circular regions, elliptical regions, and polygonal regions. 3) And constructing a common ship topological behavior set comprising four types of point-line, point-plane, line-line and line-plane according to a topological relation formed between a ship track and a navigation environment by combining a 9-intersection model (DE-9IM) based on dimension expansion. The topological behavior that a ship has in a sailing environment can each be represented as one of a set of topologies, see fig. 4.
(2) And (3) operation of spatial orientation relation: for obtaining knowledge of the orientation relationships between objects and in front of objects and the environment. The calculation steps are as follows: 1) constructing a ship-following motion coordinate system, taking a ship head line as the positive direction of an x axis, taking a ship positive transverse line as the positive direction of a y axis, 2) intersecting the x axis and the y axis to form four directional areas, and quantizing from the positive transverse direction to the head-tail direction by using a relative azimuth angle. If the ship B is in the right transverse direction of the ship A, the spatial orientation relation of the two ships A, B is as follows: the azimuth angle of the B vessel relative to the a vessel is 90 °.
(3) And (5) calculating the spatial distance relation. The distance relationship is used for describing the distance between two ship objects, and comprises quantitative description and qualitative description. Quantitative description refers to the euclidean distance between two ship objects, as shown in the following equation.
In the formula: d represents the distance between ship A and ship B, (x) A ，y A ),(x B ，y B ) Respectively, the position coordinates of vessel a and vessel B.
The qualitative description is the threshold judgment of Euclidean distance through the experience and cognitive level of a crew, and when the azimuth and the speed of two ships meet the requirement of forming collision risk, the directional description of the distance relationship between the two ships is shown as the following formula.
In the formula: d t Is a qualitative description of D, D n 、D m 、D l Is a threshold defined for the distance between ship objects.
(4) And calculating the time relation, and acquiring the time relation knowledge between the objects. The time relation is a description of the behavior and the event of the ship in a time dimension, and the time relation can be divided into a time point and a time period according to different time dimensions. Wherein the time point describes a certain moment or moment, such as the moment when the ship performs left steering, the moment when two ships collide, etc.; the time period is a time range described, such as the time when the ship is anchored at an anchorage, the time when the ship passes through a narrow water channel, and the like. In a water traffic scenario, the time point relationship, time period relationship may be described as earlier, later, intermediate, beginning, ending. Specifically, the results are shown in Table 1.
(5) And the semantic relation operation is used for acquiring semantic relation knowledge between the objects. Semantic relationships are used to describe relationships between concepts and concepts, concepts and instances, and instances. The method comprises the following specific steps: (1) defining concepts such as ship, people, environment and other elements and behavior event elements in a water traffic scene; (2) according to expert knowledge, a Resource Description Framework (RDF) is utilized to describe the semantic relationship into a triple structure (main, predicate and object) structure, a fixed domain, a value domain and a relationship of the semantic relationship and corresponding description are defined, and parts of the fixed domain, the value domain and the relationship are shown as a table to form an entity element semantic relationship knowledge base; (3) and acquiring various types of example data such as attributes corresponding to the entity concepts, inputting the various types of example data into the element semantic relation knowledge base for relation matching to acquire corresponding semantic relation knowledge.
And finding scene knowledge based on event recognition. The scene knowledge comprises: the element types, attributes and compositions of the scene, various relationships between objects and the scene, behaviors such as behaviors of the objects and events in the scene and the evolution process of the scene, and the required related elements are shown in a table. The elements of the scene and the calculation of various types of relations have been explained in the foregoing, and the scene knowledge discovery based on event recognition is explained below.
Referring to fig. 2, an example diagram of the evolution of a navigation scene based on an event trigger mechanism is shown, where the scene is the evolution from the normal navigation of a ship to the ship berthing. Firstly, identifying a sub-event 1, and finding that the ship activity 1 is a low-speed entering detection area; then identifying a sub-event 2, and finding that the ship activity 2 is that the ship stops in a detection area; sub-event 3 is then identified and the vessel activity 3 is found to be the vessel stopping in the detection zone overtime. The sub-events 1-3 are successful in succession, the occurrence of the event that the ship is parked in the detection area is triggered, so that the scene evolves, and the knowledge that the scene evolves from the normal sailing of the ship to the berthing of the ship is output.
Second, construct the device
Referring to fig. 3, a schematic diagram of a holographic navigation scene map knowledge inference device based on ontology according to an embodiment of the present invention includes: the system comprises a data preprocessing module, a scene knowledge base storage module, an object behavior information extraction module, a relation calculation module, an event recognition module and an inference knowledge output module.
The data preprocessing module is used for preprocessing the data of the sensed full-factor information and providing instantiated available information for the ontology model;
the scene knowledge base storage module is used for storing a holographic navigation scene graph knowledge base constructed based on the ontology;
the object behavior information extraction module is used for extracting the knowledge such as behavior information related to the object motion from the acquired all-element information by combining a scene knowledge base;
the relation calculation module is used for calculating the related relations such as the topological relation, the orientation relation, the distance relation, the time relation, the semantic relation and the like between the objects and the environment, and between the objects, so as to obtain the knowledge of the activities and the like of the objects;
the event identification module is used for acquiring semantic aggregation of activity knowledge and identifying sub-events occurring in a scene;
and the inference knowledge output module is used for outputting the scene evolution knowledge after the event is triggered.
The above embodiments are only used for illustrating the design idea and features of the present invention, and the purpose of the present invention is to enable those skilled in the art to understand the content of the present invention and implement the present invention accordingly, and the protection scope of the present invention is not limited to the above embodiments. Therefore, all equivalent changes and modifications made in accordance with the principles and concepts disclosed herein are intended to be included within the scope of the present invention.
Claims (9)
1. A holographic navigation scene graph knowledge inference method based on ontology is characterized in that: the method comprises the following steps:
s0: the method comprises the steps of setting up a holographic navigation scene graph knowledge inference device based on a body, wherein the holographic navigation scene graph knowledge inference device comprises a data preprocessing module, a scene knowledge base storage module, an object behavior information extraction module, a relation calculation module, an event recognition module and an inference knowledge output module;
s1: according to the understanding of a driver on objects and behaviors in a navigation scene, the space-time behaviors of the traffic objects are divided into behaviors, processes and events of the traffic objects by combining the space-time scales of ship behaviors, and semantic features of the space-time behaviors of the traffic objects under different space-time scales are represented respectively; according to the motion characteristics, the self attribute characteristics and the topological characteristics of the environment of the traffic object, respectively defining the hierarchical relationship of the time-space behaviors of the traffic object as microscopic behaviors, mesoscopic behaviors and macroscopic behaviors; in semantic cognition, the behavior of a traffic object corresponds to a microscopic behavior, the process corresponds to a mesoscopic behavior, and the event corresponds to a macroscopic behavior;
s2: carrying out structured ontology modeling on the space-time behaviors of the traffic objects, and defining concepts and attributes of the traffic objects, including behaviors, processes and events, in the ontology model;
s3: forming a scene ontology knowledge base according to the ontology model, and representing the interaction of the space-time behaviors of the objects in the sub-scene by adopting a relation calculation module;
s4: acquiring static data, dynamic data and environmental data of an object related to a scene and inputting the static data, the dynamic data and the environmental data into a scene ontology knowledge base; extracting behavior information and process information of the traffic object according to whether the behavior of the object changes in the two time slices, wherein the behavior does not change in the behavior information and the behavior changes in the process information; performing semantic aggregation on a plurality of processes to obtain an event;
s5: and (3) representing the change of the whole macroscopic scene through an event trigger mechanism, and reasoning and outputting the evolved scene knowledge.
2. The ontology-based holographic navigation scene graph knowledge inference method of claim 1, wherein: in the above-mentioned step S1, the step,
the microscopic behaviors represent the change of the attributes and the relations of the traffic objects in the space-time dimension and are basic behavior units for describing the traffic objects;
the mesoscopic behaviors represent the behaviors of the traffic objects in a larger time range and space range, and are the aggregation of microscopic behaviors;
macroscopic behavior represents the behavior of traffic objects over a large temporal and spatial extent; the behaviors of the traffic object are unchanged from a certain scale, and all the behaviors represent macroscopic behaviors; the scale is the critical scale for macroscopic behavior.
3. The ontology-based holographic navigation scene graph knowledge inference method of claim 1, wherein: in the step S1, the semantic multi-scale modeling of the spatiotemporal behavior of the traffic object specifically includes:
modeling behaviors as changes of attribute elements and relation elements of the traffic object in time and space dimensions, including motion characteristic attribute changes, self characteristic attribute changes and topological relation changes;
modeling the process as a persistence of behavior over a period of time with the object attribute behavior and the relationship behavior unchanged;
an event is modeled as the evolution of the behavior and processes of a traffic object and its logical and temporal relationships over space-time, the event comprising at least two processes.
4. The ontology-based holographic navigation scene graph knowledge inference method of claim 1, wherein: in the step S2, the specific steps are as follows:
s21: let Object be the subject Object of space-time behavior occurrence, Object Attribution Object being a property element of the subject Object Relation Is a relationship element of the subject Object, f (Object) Attribution ,Object Relation ) Is a comprehensive expression of the attribute and relationship of the main object, t is the time of occurrence of the spatio-temporal Behavior, and Behavior is the spatio-temporal object lineIs the ith behavior in the space-time object, Process is the jth Process in the space-time object behavior, Event k Is k processes in the spatio-temporal object behavior; the ontological concepts of the behavior, processes and events of the traffic object are modeled as follows:
the behavior is
The process is that
The event is
S22: defining an ontology attribute element of the traffic object, wherein the ontology attribute element comprises an object attribute and a data attribute;
the object attribute is used for representing the relation between the classes in the ontology and is the key for recognizing the ship behavior; the object properties include:
HasTraj represents the attribution relationship between the ship and the track, Domain is ship, and Ranges is track;
the compises represents the inclusion relationship between track segments and sub-tracks, Domain is track, and Ranges is track;
occurs represents the occurrence time of the ship track segment, and comprises a sub-attribute occusBegin which represents the start time of the track segment, and a sub-attribute occusEnd which represents the end time of the track segment;
the Reflect time indicates the time of the occurrence of the ship behavior, and comprises a sub-attribute refleBegin representing the starting time of the occurrence of the behavior, and a sub-attribute refleEnd representing the ending time of the behavior;
because the behavior and the track have a one-to-one correspondence relationship, the reflexedbegin is consistent with the start time occusbegin of the track section corresponding to the behavior, and the reflexend is consistent with the end time occusend of the track section corresponding to the behavior, the characteristic is that knowledge expansion is carried out in SWRL;
before and after represent the precedence relationship of instant, and the two are in reverse of relationship; the two attributes form the basis of a time logic expression mechanism, and the complex time relation in the interval is evolved based on the simple logic relation between moments;
hasBehavior represents the relationship between ships and behaviors, Domain is ship, and Ranges is behavior;
hasTopo represents the spatial topological relation between the ship track and the navigation environment, and comprises 4 relations of point-line, point-plane, line-line and line-plane; domain is trajectory, Range is trafficRule;
follow represents the continuous relationship between ship tracks and is used for constructing the logical relationship between track segments;
reflexes represent the mapping relation between track segments and behaviors, Domain is track, Ranges is behavior, and is in inverse relation with isrefleclectedBy;
HasPoint represents the dependency relationship between track segments and track points, Domain is track, Range is Point, and the sub-attributes hasBeginPoint and hasEndPoint are included;
the data attribute represents the attribute of the ontology concept and is used for describing the state of the class in the ontology; the data attributes include:
HasSpeed represents the speed attribute of the ship;
InXSDDateTimeStamp represents UTC time of instant under a time class;
the MMSI is an identification code of the ship and is used for uniquely determining the identity of the ship;
ShipType indicates the ship's category attribute.
5. The ontology-based holographic navigation scene map knowledge inference method of claim 1, characterized in that: in step S3, according to the requirement of the scene application, the topological relation, the orientation relation, the distance relation, the time relation, and the semantic relation between the objects and the environment, and between the objects are calculated to obtain the knowledge including the activities of the objects, and the specific steps are as follows:
s31: the method comprises the following steps of calculating a topological relation for acquiring topological behavior knowledge between a traffic object and an environment, and specifically comprises the following steps:
s311: the ship is regarded as a mass point, and the ship motion track comprises ship motion track points;
s312: regarding traffic rules or infrastructure in a navigation environment of a ship as an entity in a two-dimensional space, wherein the entity comprises a point P, a line L and a plane A; performing mathematical modeling on the entity according to the spatial characteristics of the entity to form mathematical expressions comprising point positions, line segments, broken lines, circular areas, elliptical areas and polygonal areas;
s313: constructing a common ship topological behavior set comprising point-line, point-plane, line-line and line-plane according to a topological relation formed between a ship track and a navigation environment by combining a 9-intersection model DE-9IM based on dimension expansion; representing a topological behavior of a ship in a sailing environment as one of a set of topologies;
s32: the orientation relation is calculated for acquiring the orientation relation knowledge between the objects and the environment, and the method specifically comprises the following steps:
s321: constructing a ship-following motion coordinate system, taking a ship head line as the positive direction of an x axis, and taking a ship positive transverse line as the positive direction of a y axis;
s322: the x axis and the y axis are intersected to form four directional areas, and the relative azimuth angle is used for quantization from the normal direction to the head-tail direction;
s33: calculating a distance relation for describing the distance between two ship objects, wherein the distance relation comprises quantitative description and qualitative description; let D denote the distance between ship A and ship B, (x) A ，y A )，(x B ，y B ) Respectively representing the position coordinates of the ship A and the ship B, the quantitative description refers to the Euclidean distance between two ship objects:
the qualitative description is the threshold judgment of Euclidean distance through experience and cognitive level of a crew; let D t Is a qualitative description of D, D n 、D m 、D t Is a threshold defined for the distance between ship objects; when the azimuth and the speed of the two ships meet the requirementsThe orientation of the distance relationship between two vessels at risk of collision is described as:
s34: calculating a time relationship for obtaining knowledge of the time relationship between the objects; the time relation is the description of the behavior and the event of the ship on the time dimension, and the time relation is divided into a time point and a time period according to the difference of the time dimension; the time point describes a certain moment or moment; the time period describes a time range;
in the water traffic scene, the time point relation and the time period relation comprise the earlier before t 1 、before(t 1 ,t 2 ) Later than after ter t 2 、after(t 1 ,t 2 ) Between Between (t) 1 ,t 2 ) Beginning with beginnWitht 1 End in endWitht 2 ；
S35: calculating semantic relations for obtaining semantic relation knowledge between objects, including describing relations between concepts, concepts and instances, the specific steps are:
s351: defining element concepts including ships, people and environments and element concepts of behavior events in the water traffic scene;
s352: according to expert knowledge, a resource description framework is utilized to describe the semantic relation into a triple structure < main, predicate and object >, a definition domain, a value domain and a relation of the semantic relation and corresponding description are defined, and an entity element semantic relation knowledge base is formed; the method comprises the following steps:
the definition domain is a ship, the relation is happy, the value domain is an event element and is described as that a certain ship generates a certain event;
the definition domain is a ship, the relation is execute, the value domain is a behavior element, and the definition domain is described as a certain ship executing a certain behavior;
the definition domain is a ship, the relation is hasType, the value domain is a parameter attribute and is described as the type of a certain ship;
the definition domain is a ship, the relation is a hash, the value domain is a parameter attribute, and the value domain is described as the name of a certain ship;
the definition domain is a ship, the relation is hassped, the value domain is a parameter attribute and is described as the speed of a certain ship;
the defined domain is a ship, the relation is hascourse, the value domain is a parameter attribute, and the value domain is described as the course of a certain ship;
the definition domain is a ship, the relation is hazize, the value domain is a parameter attribute, and the value domain is described as the size of a certain ship;
defining a domain as a behavior element, a relation as trigger, and a value domain as a behavior element, wherein the value domain is described as a behavior element triggering a certain behavior;
the definition domain is a behavior element, the relation is cause, the value domain is an event element, and the definition domain is described as an event caused by a certain behavior;
the definition domain is a behavior element, the relationship is operatedBy, the value domain is an operator, and the description is that a certain behavior is operated by a certain person;
defining a domain as an event element, a relationship as hasType, a value domain as a parameter attribute and describing the type of a certain event;
defining a domain as an event element, a relation as trigger, a value domain as a behavior element, and describing that a certain event triggers a certain behavior;
the definition domain is an event element, the relation is cause, the value domain is an event element and is described as an event causing the event;
s353: and acquiring attributes corresponding to the entity concepts, and inputting the attributes into the element semantic relation knowledge base for relation matching to acquire corresponding semantic relation knowledge.
6. The ontology-based holographic navigation scene map knowledge inference method of claim 1, characterized in that: in step S5, the scene knowledge includes element type, attribute, and composition of the scene, various relationships between objects and objects in the scene, and between objects and the scene, and behavior, events, and scene evolution process of the objects in the scene;
in the scene evolution process type, the included elements comprise a process type and a name, a process semantic description, a time period, an included object, an included behavior, a behavior relation set, a process-process relation, a process-scene relation and a process-element relation;
in the scene element type, the included elements include element names, time, spatial features, attribute features, semantic descriptions, element relationships, element participation processes, element participation behaviors, element affiliated scenes, element-process relationships and element-scene relationships;
in the behavior type, the included elements comprise the behavior type, the name, the behavior participant, the time, the place, the behavior precursor, the behavior successor and the participant relationship;
in the state type, the included elements comprise the state type, the name, the time, the place, the state characteristics of the elements and the state of a scene; the state characteristics of the elements comprise positions, attributes and semantics; the state of the scene includes the number of elements, spatial distribution, and relationships between elements.
7. A knowledge inference device for the ontology-based holographic navigation scenario knowledge inference method of any one of claims 1 to 6, characterized in that: the system comprises a data preprocessing module, a scene knowledge base storage module, an object behavior information extraction module, a relation calculation module, an event recognition module and an inference knowledge output module;
the data preprocessing module is used for preprocessing the data of the sensed full-factor information and providing instantiated available information for the ontology model;
the scene knowledge base storage module is used for storing a holographic navigation scene graph knowledge base constructed based on the ontology;
the object behavior information extraction module is used for extracting behavior information knowledge related to object motion characteristics, self attribute characteristics and relationship change among objects from the acquired full-factor information by combining a scene knowledge base;
the relation calculation module is used for calculating the related relations such as the topological relation, the orientation relation, the distance relation, the time relation, the semantic relation and the like between the objects and the environment, between the objects, so as to obtain the knowledge of the process including the objects;
the event recognition module is used for acquiring semantic aggregation of process knowledge and recognizing sub-events occurring in a scene;
and the inference knowledge output module is used for outputting the scene evolution knowledge after the event is triggered.
8. The knowledge inference apparatus of claim 7, wherein:
adopting prot g as a rule ontology modeling tool, and adopting a pellet inference machine as an inference knowledge output module;
the ontology modeling tool prot g and the pellet inference engine both support OWL language and SWRL language;
the OWL language is used for describing rule knowledge; the SWRL language is a regular knowledge inference language.
9. A computer storage medium, characterized in that: stored with a computer program executable by a computer processor, the computer program performing a method of ontology-based holographic navigation scene knowledge inference according to any of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210597124.7A CN114926611A (en) | 2022-05-30 | 2022-05-30 | Holographic navigation scene graph knowledge inference method and device based on ontology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210597124.7A CN114926611A (en) | 2022-05-30 | 2022-05-30 | Holographic navigation scene graph knowledge inference method and device based on ontology |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114926611A true CN114926611A (en) | 2022-08-19 |
Family
ID=82812297
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210597124.7A Pending CN114926611A (en) | 2022-05-30 | 2022-05-30 | Holographic navigation scene graph knowledge inference method and device based on ontology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114926611A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116049501A (en) * | 2023-01-30 | 2023-05-02 | 北京化工大学 | Method for generating natural language description of spatial relation of spatial scene |
| CN116562172A (en) * | 2023-07-07 | 2023-08-08 | 中国人民解放军国防科技大学 | Geographical scene time deduction method, device and equipment for space-time narrative |
-
2022
- 2022-05-30 CN CN202210597124.7A patent/CN114926611A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116049501A (en) * | 2023-01-30 | 2023-05-02 | 北京化工大学 | Method for generating natural language description of spatial relation of spatial scene |
| CN116562172A (en) * | 2023-07-07 | 2023-08-08 | 中国人民解放军国防科技大学 | Geographical scene time deduction method, device and equipment for space-time narrative |
| CN116562172B (en) * | 2023-07-07 | 2023-09-15 | 中国人民解放军国防科技大学 | Geographical scene time deduction method, device and equipment for space-time narrative |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ding et al. | A survey on safety-critical driving scenario generation—A methodological perspective | |
| Käfer et al. | Recognition of situation classes at road intersections | |
| CN114926611A (en) | Holographic navigation scene graph knowledge inference method and device based on ontology | |
| Buxton et al. | Visual surveillance in a dynamic and uncertain world | |
| Fernyhough et al. | Constructing qualitative event models automatically from video input | |
| Rota et al. | Activity recognition from video sequences using declarative models | |
| Zhang et al. | Vessel trajectory prediction in maritime transportation: Current approaches and beyond | |
| Mascaro et al. | Learning abnormal vessel behaviour from AIS data with Bayesian networks at two time scales | |
| CN107730903A (en) | Parking offense and the car vision detection system that casts anchor based on depth convolutional neural networks | |
| US20240043026A1 (en) | Performance testing for trajectory planners | |
| Kolekar et al. | Behavior prediction of traffic actors for intelligent vehicle using artificial intelligence techniques: A review | |
| Weidl et al. | Early and accurate recognition of highway traffic maneuvers considering real world application: A novel framework using Bayesian networks | |
| CN113867367B (en) | Processing method and device for test scene and computer program product | |
| Liu et al. | Dynamic image sequence analysis using fuzzy measures | |
| CN115018215B (en) | Population residence prediction method, system and medium based on multi-modal cognitive atlas | |
| CN114492544B (en) | Model training method and device and traffic incident occurrence probability evaluation method and device | |
| Jakob et al. | Employing agents to improve the security of international maritime transport | |
| Burns et al. | Temporal abstraction in Bayesian networks | |
| Szwed et al. | Object tracking and video event recognition with fuzzy semantic petri nets | |
| Song et al. | Data-driven cognitive modeling and semantic reasoning of ship behavior | |
| Fournier et al. | TRANS: a tractable role-based agent prototype for concurrent navigation systems | |
| US20240112044A1 (en) | Methods and systems for ontology construction with ai-mediated crowdsourcing and concept mining for high-level activity understanding | |
| US20240112473A1 (en) | Object trajectory clustering with hybrid reasoning for machine learning | |
| Dee | Explaining visible behaviour | |
| Zhou et al. | A Multimodal Trajectory Prediction Method for Pedestrian Crossing Considering Pedestrian Motion State |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |